Abstract
AIM: This study aimed to identify optimal methods of acceleration analysis to accurately detect dual-task-related changes in walking sway. METHODS: Twenty-six healthy adults participated in this study, undergoing various cognitive dual-task conditions while walking. Accelerometers were attached to the lower back to record center-of-mass (COM) acceleration in anterior-posterior (AP) and mediolateral (ML) directions. Data analysis involved multiple computation methods applied to the acceleration data, with comparisons made using different frequency cutoffs and mean referencing. RESULTS: Analysis revealed significant effects of dual-tasking on walking sway, particularly in AP and combined directions. A 3.5Hz low-pass filter and mean referencing were found effective in capturing these changes. Computation methods, such as root mean square (which quantifies the variability of COM acceleration) and normalized path length (which measures the distance traveled by the COM over time), showed sensitivity to detect changes in sway amplitude during dual-task conditions. CONCLUSION: The findings highlight the importance of considering both AP and ML sway in dual-task assessments. Furthermore, the choice of computation method, frequency cutoff, and mean referencing impacted the sensitivity to detect changes in walking sway during dual-tasking. These findings suggest that a 3.5Hz low-pass filter with mean referencing can enhance the sensitivity of dual-task assessments, which could be valuable for clinical balance evaluations or rehabilitation monitoring.